Back to skill

Security audit

ARCHIV

Security checks for vulnerabilities and agentic risk

Overview

This Roon music-control skill needs Review because it bundles unrelated long-term memory, credential locations, and backup/upload guidance beyond music playback.

Install only after removing MEMORY.md or ensuring it cannot be imported as trusted memory. Also narrow activation to explicit Roon requests, require confirmation for queue-clearing or volume changes, replace hardcoded host and zone IDs with local configuration, and protect the Roon API with authentication or network isolation.

Vulnerability Patterns
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Findings (5)

T02 · Agent Memory Poisoning

Error
Location
MEMORY.md:60
Finding

Unrelated Persistent Instructions Poison Agent Memory and Future Behavior

Content
View full analysis
1 && tolower($0) ~ /suchbegriff/i {print $0}' /root/clawd/docs/imdb_ratings.csv | head -5` (sucht in ALLEN Feldern case-insensitive) * Trakt: `trakt-cli history --limit 200 | grep -i suchbegriff || true` * **Prozess**: IMMER zue ...[truncated 1690 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
MEMORY.md:3
Finding

Exposure of Credential Locations and Private Host Resources

Content
View full analysis
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
MEMORY.md:90
Finding

Unrelated Secret-Bearing External API Request Exceeds Skill Privileges

Content
View full analysis
Remediation
View remediation

T06 · System Persistence

Error
Location
MEMORY.md:7
Finding

Persistent Memory Export and Local-Path Transfer Bypass

Content
View full analysis
Remediation
View remediation

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:16
Finding

Roon Control API Uses Unauthenticated Plaintext HTTP

Content
View full analysis
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (9)

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · MEMORY.md (reported line 10)May include surrounding context.

md
## 💾 **Git-Backup** (seit 13.02.2026)
- **Repo:** https://github.com/WegoW/Openclaw.git
- **Hourly Cron:** MEMORY.md + memory/ auto commit/push (`scripts/backup.sh`)
- **Security:** NIE API-Keys in memory/*.md oder MEMORY.md speichern - immer .env + .gitignore + git-filter-repo für History-Cleanup bei Leaks

## 🎵 **Musik-Präferenz** (seit 02.02.2026)
- **Tool:** `mcporter call unified-hifi-control.hifi_zones`, `.hifi_now_playing zone_id=...`, `.hifi_play(...)` für Roon

Vague Triggers

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill description explicitly tells the agent to trigger on broad, ordinary phrases like 'put something on', 'skip this', and 'turn it up', and claims it can 'act immediately' with full knowledge of zones and library. This creates a strong risk of unintended invocation and unauthorized playback or device control from ambiguous conversation, especially because the skill performs real side effects.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

Line L87 specifies a German-only fallback TTS voice (de-DE-KillianNeural) with no indication that the user can choose another language or opt in to German output. This creates a natural-language locale constraint that may conflict with organizational language-choice expectations.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
86% confidence
Finding

The skill instructs use of exec+curl to contact an external API endpoint and include an API key from secure storage. Even though the endpoint is legitimate, this creates a defined exfiltration path to a third party and encourages direct shell-based network access with secrets, which can be abused or mishandled in a broader agent workflow.

Content

Scanner excerpt · MEMORY.md (reported line 91)May include surrounding context.

md
**Source:** User-Anweisung Walter Wego (20.02.2026)

## 🔊 **ElevenLabs Credits** (seit 20.02.2026)
- **Abruf:** `exec` mit `curl -X GET 'https://api.elevenlabs.io/v1/user' -H 'xi-api-key: [SECURE_STORAGE]' -H 'accept: application/json'`
- **Parse:** `subscription.character_limit - subscription.character_count` = verbleibend
- **Anzeige:** Nur "**Verbleibend:** X / Y Zeichen (Free Tier, monatlich). 🤓"

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 29)May include surrounding context.

GET:

bash
curl -s 'http://roonext2.home:3001/api/status'

POST:

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 93)May include surrounding context.

Supports the same action strings: Play Now, Queue, Add Next, Start Radio.

bash
curl -X POST http://roonext2.home:3001/api/play-album \
  -H "Content-Type: application/json" \
  -d '{
    "zone_id": "YOUR_ZONE_ID",

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The documentation states that 'Play Now' starts playback and clears the existing queue, but it does not require warning or confirmation before using this destructive action. In a music-control skill, silently replacing the current queue can disrupt user activity and cause unintended loss of the listening session state.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 145)May include surrounding context.

Set shuffle to true to enable, false to disable.

bash
curl -s -X POST http://roonext2.home3001/api/shuffle \
  -H 'Content-Type: application/json' \
  -d '{"zone_id": "1601dcef8115529daf4cd6807753971fae3e", "shuffle": true}'

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · SKILL.md (reported line 208)May include surrounding context.

md
for track in "${tracks[@]}"; do
  query="${track%%|*}"
  action="${track##*|}"
  curl -s -X POST "$API" \
    -H 'Content-Type: application/json' \
    -d "{\"zone_id\":\"$ZONE\",\"query\":\"$query\",\"type\":\"Tracks\",\"action\":\"$action\"}"
  sleep 2